Testing using Privileged Information by Adapting Features with Statistical Dependence

TL;DR

This paper proposes a method to improve predictions by leveraging additional features at test time through statistical dependence, without retraining or access to the original prediction function.

Contribution

It introduces a novel approach that exploits statistical dependence and manifold denoising to enhance predictions using privileged information at test time.

Findings

Improved visual attribute ranking results.

Effective enhancement without retraining or original model access.

Method applicable to real-world prediction tasks.

Abstract

Given an imperfect predictor, we exploit additional features at test time to improve the predictions made, without retraining and without knowledge of the prediction function. This scenario arises if training labels or data are proprietary, restricted, or no longer available, or if training itself is prohibitively expensive. We assume that the additional features are useful if they exhibit strong statistical dependence to the underlying perfect predictor. Then, we empirically estimate and strengthen the statistical dependence between the initial noisy predictor and the additional features via manifold denoising. As an example, we show that this approach leads to improvement in real-world visual attribute ranking. Project webpage: http://www.jamestompkin.com/tupi